The attached plots show a length noise injection into the PMC using TFIN. The first plot is calibrated in Volts for the HV_MON, whereas the second plot is calibrated in displacement. The coupling coefficient at 1 kHz is about 3.5 x 10-8 between the PZT and DARM length.
The reference traces 0 and 1 are without injection, whereas reference traces 2, 3 and 4 are from before the changes on the PSL table on Tuesday.
A measurement done at the very beginning of the run had about half the coupling—indicating that the coupling depends on thermal lensing.
The attached plot shows frequency and intensity noise measured by the IMC/REFL and second loop ISS sensors.
There is coherence above 1 kHz with intensity noise. Using the intensity noise coupling TF from alogs 30274 or 29926, one can conclude that it doesn't couple through intensity noise. The intensity noise is at 10-8 RIN/rtHz at 1 kHz. With a coupling of 10-13 m/RIN, we are far below the DARM noise.
The frequency noise produced by the PMC length noise is of the order of 1 Hz/rtHz at 1 kHz. It will be suppressed by the FSS, IMC and REFL. As such it is much too small to be responsible for the coupling directly. There is however, coherence with frequency noise as seen by REFL_A_RF9_I in the 1 kHz region. Assuming the REFL control signal is dominated by IMC sensing noise which seems to be around 2 x 10-4 Hz/rtHz at 1 kHz for 25 W input, see alog 31138, and using the noise coupling from alog 31176, we get a number around 10-19 m/rtHz at 1 kHz. So, it is conceivable that the PMC noise produces error point offsets in the REFL servo which in turn propagate to DARM as frequency noise.
We saw an increase of the PMC length noise coupling to DARM, when we misaligned SRM by 20 µrad. According T0900142 the requirement for SRM is 2.5 µrad which puts the required 3x10-6 /rtHz at 1 kHz with a safety factor of 10. So, for 20 µrad, a jitter of 7x10-6/rtHz at 1kHz is at the DARM noise level at full sensitivity. Our jitter was roughly 5x10-6/rtHz and it made a difference at the 1x10-19 level. Maybe somewhat higher than expected but close.
Looking at the IMC WFS signal we can clearly see the PMC length noise injection. If we take the 260 Hz periscope peak as a reference, this doesn't explain the coupling to DARM. What is surprising is that even without length noise injection (REF traces), the coherence between IMC WFS and PMC HVMon is large at frequencies between the jitter peaks.
This measurement was repeated with just the IMC locked at 25 W and ISS second loop enageged. This did not change the HV MON coherence with the IMC WFS nor the coherence with MC_F.